Nguyễn Công Phương
SIGNAL PROCESSING
Transform Analysis of LTI Systems
Contents
I. Introduction
II. Discrete – Time Signals and Systems
III. The z – Transform
IV. Fourier Representation of Signals
V. Transform Analysis of LTI Systems
VI. Sampling of Continuous – Time Signals
VII.The Discrete Fourier Transform
VIII.Structures for Discrete – Time Systems
IX. Design of FIR Filters
X. Design of IIR Filters
XI. Random Signal Processing
sites.google.com/site/ncpdhbkhn
2
Transform Analysis
of LTI Systems
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
Sinusoidal Response of LTI Systems
Response of LTI Systems in the Frequency Domain
Distortion of Signals Passing through LTI Systems
Ideal and Practical Filters
Frequency Response for Rational System Functions
Dependency of Frequency Response on Poles and Zeros
Design of Simple Filters by Pole – Zero Placement
Relationship between Magnitude and Phase Responses
Allpass Systems
Invertibility and Minimum – Phase Systems
Transform Analysis of Continuous – Time Systems
sites.google.com/site/ncpdhbkhn
3
Sinusoidal Response of LTI
Systems (1)
x[ n ] = z → y[n ] = H ( z ) z , all n;
H ( z ) = ∑ h[ k ]z
∞
n
n
−k
k =−∞
z = e jω → x[n] = e jωn → y[n] = H (e jω )e jωn , all n
jω
H (e ) = H ( z ) z =e jω =
jω
jω
H (e ) = H (e ) e
x[n ] = Ae
j ( ω n +φ )
j∠H ( e jω )
∞
− jωk
h
[
k
]
e
∑
k =−∞
= H R (e jω ) + jH I (e jω )
jω
→ y[n] = A H (e ) e
j [ωn +φ +∠H ( e jω )]
The response of a stable LTI system to a complex exponential sequence
is a complex exponential sequence with the same frequency,
only the amplitude and phase are changed by the system.
sites.google.com/site/ncpdhbkhn
4
Sinusoidal Response of LTI
Systems (2)
Ax jφx jωn Ax − jφx − jωn
x[n ] = Ax cos(ωn + φx ) =
e e + e e
2
2
x[n ] = Ae
j ( ω n +φ )
jω
→ y[n] = A H (e ) e
j [ωn +φ +∠H ( e jω )]
Ax jφx jωn
Ax
jφx j [ ωn +∠H ( e jω )]
jω
x1[n ] =
e e → y1[n ] =
H (e ) e e
2
2
Ax − jφx − jωn
Ax
− jφ x j [ − ωn +∠H ( e − jω )]
− jω
x2 [n] =
e e
→ y2 [ n ] =
H (e ) e e
2
2
Ax
Ax
j [ω n +φ x +∠H ( e jω )]
j [ − ωn −φx +∠H ( e − jω )]
jω
− jω
→ y[ n ] =
H (e ) e
+
H (e ) e
2
2
sites.google.com/site/ncpdhbkhn
5
Sinusoidal Response of LTI
Systems (3)
Ax
Ax
j[ ωn +φ x +∠H ( e jω )]
j[ − ωn −φ x +∠H ( e − jω )]
jω
− jω
y[ n ] =
H (e ) e
+
H (e ) e
2
2
H (e − jω ) = H (e jω ) ;
∠H (e− jω ) = −∠H (e− jω )
Ax
Ax
j [ ωn +φx +∠H ( e jω )]
− j [ ωn +φx +∠H ( e jω )]
jω
jω
→ y[ n ] =
H (e ) e
+
H (e ) e
2
2
Ax
j [ ωn +φx +∠H ( e jω )]
− j [ ωn +φ x +∠H ( e jω )]
jω
=
H (e ) {e
+e
}
2
= Ax H (e jω ) cos ωn + φx + ∠H (e jω )
x[ n ] = Ax cos(ωn + φx ) → y[n ] = Ay cos(ωn + φ y )
Ay = Ax H ( e jω )
( magnitude response/gain )
φ y = ∠H (e jω ) + φx (phase response)
sites.google.com/site/ncpdhbkhn
6
Ex. 1
Sinusoidal Response of LTI
Systems (4)
Find the frequency response of the system y[n] = ay[n – 1] + bx[n],
–1 < a < 1.
x[n] = e jωn → y[n] = H (e jω )e jωn
y[n − 1] = H (e jω )e jω ( n −1)
→ H ( e jω )e jωn = aH ( e jω )e jω ( n −1) + be jωn
b
y[ n] = ay[ n − 1] + bx[ n] → H (e jω ) =
1 − ae − jω
e − jω = cos ω − j sin ω
→ H (e jω ) = b
1 − cos ω
a sin ω
−
jb
1 − 2a cos ω + a 2
1 − 2a cos ω + a 2
b
jω
H (e ) =
2
1
−
2
a
cos
ω
+
a
→
∠H (e jω ) = atan −a sin ω
1 − a cos ω
sites.google.com/site/ncpdhbkhn
7
Sinusoidal Response of LTI
Systems (5)
Ex. 1
Find the frequency response of the system y[n] = ay[n – 1] + bx[n],
b
H (e jω ) =
1 − 2a cos ω + a 2
∠H (e jω ) = atan
;
–1 < a < 1.
−a sin ω
1 − a cos ω
Magnitude
2
1.5
1
0.5
-3
-2
-1
0
omega (rad)
1
2
3
-3
-2
-1
0
omega (rad)
1
2
3
0.6
0.4
Phase
0.2
0
-0.2
-0.4
sites.google.com/site/ncpdhbkhn
8
Sinusoidal Response of LTI
Systems (6)
x[n] = e jωn → y[n ] = H (e jω )e jωn , all n
x[n] = e
jω n
n
n
k =0
k =0
u[n] → y[n ] = ∑ h[k ]x[n − k ] = ∑ h[k ]e jω ( n −k )
∞
∞
− jω k jω n
− jωk jωn
= ∑ h[k ]e
e
−
h
[
k
]
e
∑
e
k =0
k =n +1
jω
= H (e )e
jω n
y steady state [ n ]
∞
− jωk jωn
− ∑ h[k ]e
e
k = n +1
ytransient response [ n ]
ytr [n] ≤
∞
∑
k = n +1
∞
h[k ] ≤ ∑ h[k ] < ∞
k =0
lim y[n] = H (e jω )e jωn = yss [n]
n →∞
sites.google.com/site/ncpdhbkhn
9
Ex. 2
Sinusoidal Response of LTI
Systems (7)
Given a system h[n] = 0.7nu[n] and an input x[n] = cos(0.03πn)u[n], find the output?
y[ n] = h[ n]* x[n ] → Y ( z ) = H ( z ) X ( z )
H ( z) =
1
,
1 − 0.7 z −1
z > 0.7
1 − cos(0.03π ) z −1
X ( z) =
,
−1
−2
1 − 2 cos(0.03π ) z + z
z >1
1 − cos(0.03π ) z −1
→ Y ( z) =
(1 − 0.7 z −1 )[1 − 2 cos(0.03π ) z −1 + z −2 ]
1 − 2 cos(0.03π )z −1 + z −2 = 0
∆ = 4 cos2 α − 4 = − 4 sin 2 α
−1
→ ( z )1,2
2 cos α ± −4 sin 2 a
=
= cos α ± j sin α
2
= e ± jα = e ± j 0.03π
→ 1 − 2 cos(0.03π ) z −1 + z −2 = ( z −1 − e j 0.03π )( z −1 − e − j 0.03π ) = (1 − e j 0.03π z −1 )(1 − e − j 0.03π z −1 )
1 − cos(0.03π ) z −1
→ Y ( z) =
(1 − 0.7 z −1 )(1 − e j 0.03π z −1 )(1 − e − j 0.03π z −1 )
sites.google.com/site/ncpdhbkhn
10
Ex. 2
Sinusoidal Response of LTI
Systems (8)
Given a system h[n] = 0.7nu[n] and an input x[n] = cos(0.03πn)u[n], find the output?
1 − cos(0.03π ) z −1
K1
K2
K3
Y ( z) =
=
+
+
(1 − 0.7 z −1 )(1 − e j 0.03π z −1 )(1 − e − j 0.03π z −1 ) 1 − 0.7 z −1 1 − e j 0.03π z −1 1 − e − j 0.03π z −1
→ 1 − cos(0.03π ) z −1 = K1 (1 − e j 0.03π z −1 )(1 − e − j 0.03π z −1 ) + K 2 (1 − 0.7 z −1 )(1 − e − j 0.03π z −1 ) +
+ K3 (1 − 0.7 z −1 )(1 − e j 0.03π z −1 )
z −1 = 0.7 −1 → 1 − cos(0.03π )0.7 −1 = K1 (1 − e j 0.03π 0.7 −1 )(1 − e − j 0.03π 0.7 −1 ) +
+ K 2 (1 − 0.7 × 0.7 −1 )(1 − e − j 0.03π 0.7 −1 ) + K 3 (1 − 0.7 × 0.7 −1 )(1 − e j 0.03π 0.7 −1 )
1 − 0.7 −1 cos(0.03π )
→ K1 =
(1 − 0.7 −1 e j 0.03π )(1 − 2 e − j 0.03π )
1 − 0.7−1 cos(0.03π )
1 − 0.7−1 cos(0.03π )
=
=
−1 j 0.03π
− 1 − j 0.03π
−2
1 − 0.7 e
− 0 .7 e
+ 0 .7
1 + 0.7 −2 − 0.7 −1( e j 0.03π + e − j 0.03π )
e jθ = cos θ + j sin θ
e − jθ = cos θ − j sin θ
→ e j 0.03π + e− j 0.03π = 2 cos(0.03π )
1 − 0.7 −1 cos(0.03π )
→ K1 =
= −2.1504
1 + 0.7 −2 − 0.7 −12 cos(0.03π )
sites.google.com/site/ncpdhbkhn
11
Ex. 2
Sinusoidal Response of LTI
Systems (9)
Given a system h[n] = 0.7nu[n] and an input x[n] = cos(0.03πn)u[n], find the output?
1 − cos(0.02π ) z −1
−2.1504
K2
K3
Y ( z) =
=
+
+
(1 − 0.5z −1 )(1 − e j 0.02π z −1 )(1 − e − j 0.02π z −1 ) 1 − 0.7 z −1 1 − e j 0.03π z −1 1 − e − j 0.03π z −1
→ 1 − cos(0.03π ) z −1 = −2.1504(1 − e j 0.03π z −1 )(1 − e − j 0.03π z −1 ) + K 2 (1 − 0.7 z −1 )(1 − e − j 0.03π z −1 ) +
+ K 3 (1 − 0.7 z −1 )(1 − e j 0.03π z −1 )
z −1 = e − j 0.03π → 1 − cos(0.03π )e − j 0.03π = K 2 (1 − 0.7e − j 0.03π )(1 − e − j 0.03π e − j 0.03π )
1 − e − j 0.03π cos(0.03π )
→ K2 =
= 1.6120∠ − 0.2140
(1 − 0.7 e − j 0.03π )(1 − e − j 2×0.03π )
z −1 = e j 0.03π → 1 − cos(0.03π )e j 0.03π = K 3 (1 − 0.7e j 0.03π )(1 − e j 0.03π e j 0.03π )
1 − e − j 0.03π cos(0.03π )
→ K3 =
= 1.6120∠0.2140
(1 − 0.7e j 0.03π )(1 − e j 2×0.03π )
→ Y ( z) =
−2.1504 1.6120∠ − 0.2140 1.6120∠0.2140
+
+
1 − 0.7 z −1
1 − e j 0.03π z −1
1 − e − j 0.03π z −1
sites.google.com/site/ncpdhbkhn
12
Ex. 2
Sinusoidal Response of LTI
Systems (10)
Given a system h[n] = 0.7nu[n] and an input x[n] = cos(0.03πn)u[n], find the output?
Y ( z) =
−2.1504 1.6120∠ − 0.2140 1.6120∠0.2140
+
+
1 − 0.7 z −1
1 − e j 0.03π z −1
1 − e − j 0.03π z −1
X ( z) =
K
n
→
x
[
n
]
=
Kp
u[n ]
−1
1 − pz
→ y[n ] = −2.1504(0.7)n u[ n] + (1.6120∠ − 0.2140)(e j 0.03π ) n u[ n] + (1.6120∠0.2140)( e − j 0.03π )n u[n ]
= [ −2.1504(0.7)n + 1.6120e − j 0.2140 ( e j 0.03π ) n + 1.6120e j 0.2140 (e j 0.03π ) n ]u[n ]
= {−2.1504(0.7)n + 1.6120[ e j ( 0.03π n−0.2140 ) + e −( 0.03π n −0.2140) ]}u[ n]
= −2.1504(0.7) n u[n] + 1.6120 × 2 cos(0.03π n − 0.2140)u[ n]
ytransient response [ n ]
ysteady state [ n ]
sites.google.com/site/ncpdhbkhn
13
Sinusoidal Response of LTI
Systems (11)
Ex. 2
Given a system h[n] = 0.7nu[n] and an input x[n] = cos(0.03πn)u[n], find the output?
y[ n] = −2.1504(0.7)n u[ n] + 1.6120 × 2 cos(0.03π n − 0.2140)u[ n]
ytransient response [ n ]
y steady state [ n ]
4
x[n]
y[n]
3
2
1
0
-1
-2
-3
-4
0
20
40
60
80
Time index (n)
sites.google.com/site/ncpdhbkhn
100
120
140
160
14
Transform Analysis
of LTI Systems
1.
2.
Sinusoidal Response of LTI Systems
Response of LTI Systems in the Frequency Domain
a)
b)
c)
3.
4.
5.
6.
7.
8.
9.
10.
11.
Response to Periodic Inputs
Response to Aperiodic Inputs
Power Gain
Distortion of Signals Passing through LTI Systems
Ideal and Practical Filters
Frequency Response for Rational System Functions
Dependency of Frequency Response on Poles and Zeros
Design of Simple Filters by Pole – Zero Placement
Relationship between Magnitude and Phase Responses
Allpass Systems
Invertibility and Minimum – Phase Systems
Transform Analysis of Continuous – Time Systems
sites.google.com/site/ncpdhbkhn
15
Response to Periodic Inputs (1)
N −1
x[n ] = ∑ ck( x )e
x[ n] = x[ n + N ],
j
2π
kn
N
k =0
2π
kn
N
j
e
N −1
x[n] = ∑ c e
k =0
( x)
k
c
c
= H (e
Py =
1
N
N −1
∑
n=0
2π
kn
N
j
N −1
2π
k
N
j
2π
k
N
(x)
k
) c
2π
k
N
→ ∑ c H (e
k =0
= H (e
(y)
k
(y)
k
j
→ y[n ] = H (e
j
(x)
k
N −1
2π
k
N
2π
kn
N
)e
∠c
( y)
k
= ∠H ( e
N −1
y[n] = ∑ ck( y ) = ∑ H (e
2
k =0
j
2π
kn
N
= y[n]
−∞< k < ∞
)ck( x ) ,
,
j
)e
j
2
k =0
sites.google.com/site/ncpdhbkhn
j
2π
k
N
j
2π
k
N
) + ∠ck( x )
2
) ck( x )
2
16
Response to Periodic Inputs (2)
Ex.
1,
0,
Given a periodic sequence x[ n] =
0≤n< N
N ≤n
with fundamental period M
and a system function h[n] = anu[n]. Find the output sequence?
sites.google.com/site/ncpdhbkhn
17
Transform Analysis
of LTI Systems
1.
2.
Sinusoidal Response of LTI Systems
Response of LTI Systems in the Frequency Domain
a)
b)
c)
3.
4.
5.
6.
7.
8.
9.
10.
11.
Response to Periodic Inputs
Response to Aperiodic Inputs
Power Gain
Distortion of Signals Passing through LTI Systems
Ideal and Practical Filters
Frequency Response for Rational System Functions
Dependency of Frequency Response on Poles and Zeros
Design of Simple Filters by Pole – Zero Placement
Relationship between Magnitude and Phase Responses
Allpass Systems
Invertibility and Minimum – Phase Systems
Transform Analysis of Continuous – Time Systems
sites.google.com/site/ncpdhbkhn
18
Response to Aperiodic Inputs
Y (e jω ) = H (e jω ) X ( e jω )
Y (e jω ) = H (e jω ) X (e jω )
→
jω
jω
jω
∠Y (e ) = ∠H (e ) + ∠X (e )
sites.google.com/site/ncpdhbkhn
19
Power Gain
jω
Gain in dB = H (e )
jω
dB
= 10 log10 H (e )
Y (e jω ) = H (e jω ) X (e jω ) → Y (e jω )
dB
= H (e jω )
sites.google.com/site/ncpdhbkhn
dB
2
+ X (e jω )
dB
20
Transform Analysis
of LTI Systems
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
Sinusoidal Response of LTI Systems
Response of LTI Systems in the Frequency Domain
Distortion of Signals Passing through LTI Systems
Ideal and Practical Filters
Frequency Response for Rational System Functions
Dependency of Frequency Response on Poles and Zeros
Design of Simple Filters by Pole – Zero Placement
Relationship between Magnitude and Phase Responses
Allpass Systems
Invertibility and Minimum – Phase Systems
Transform Analysis of Continuous – Time Systems
sites.google.com/site/ncpdhbkhn
21
Distortion of Signals Passing
through LTI Systems (1)
A system has distortionless response if the input signal x[n]
and the output signal y[n] have the same “shape”
y[n ] = Gx[ n − nd ],
G >0
→ Y (e jω ) = Ge − jωnd X (e jω )
Y ( e jω )
− jωnd
→ H (e ) =
=
Ge
X ( e jω )
jω
H (e jω ) = G
→
jω
∠H (e ) = −ωnd
sites.google.com/site/ncpdhbkhn
22
Distortion of Signals Passing
through LTI Systems (2)
y[n ] = Ax H (e jω ) cos ωn + φx + ∠H (e jω )
jω
φ
∠
H
(
e
)
jω
x
= Ax H (e ) cos ω n + +
ω
ω
τ phase delay (ω ) = −
∠H (e jω )
ω
d Ψ(ω )
τ group delay (ω ) = −
dω
x[n ] = s[n]cos ωc n → y[n ] ≈ H (e jωc ) s[n − τ gd (ωc )]cos{ωc [n − τ pd (ωc )]}
sites.google.com/site/ncpdhbkhn
23
Transform Analysis
of LTI Systems
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
Sinusoidal Response of LTI Systems
Response of LTI Systems in the Frequency Domain
Distortion of Signals Passing through LTI Systems
Ideal and Practical Filters
Frequency Response for Rational System Functions
Dependency of Frequency Response on Poles and Zeros
Design of Simple Filters by Pole – Zero Placement
Relationship between Magnitude and Phase Responses
Allpass Systems
Invertibility and Minimum – Phase Systems
Transform Analysis of Continuous – Time Systems
sites.google.com/site/ncpdhbkhn
24
Ideal and Practical Filters (1)
H lp ( e jω )
1
1
−π
−ωc 0 ωc
π
ω
−π −ωc
Lowpass
Highpass
Bandpass
Bandstop
H bp ( e jω )
0
ωl
ωu
π
ω
−π
−ωu
sites.google.com/site/ncpdhbkhn
ωc π
0
1
1
−π
−ωu −ωl
H hp ( e jω )
ω
H bs (e jω )
−ωl 0 ωl
ωu
π
ω
25